【作 者】：
吴军其， 吴飞燕， 张萌萌， 戴新菊， 张 影
【关 键 词 】：
多模态学习分析； 智慧课堂； 协作学习； 学习投入度； 模型
【栏 目】：
学习环境与资源
【中文摘要】：
在协作学习中，学习投入是影响学生学业成功和教育教学质量的重要因素，对其有效测评能为教师研判协作活动提供依据。人工智能和智能感知技术的发展催生了智慧课堂，学习投入测评方式也由单一维度向多模态融合分析转变。文章基于多模态学习分析理论，结合智慧课堂协作学习的特点，探讨技术与学习测评的深度融合，从行为、认知和情感三个维度分析学习投入，构建智慧课堂协作学习投入度分析模型并在实际课堂中应用。实践发现来自智慧课堂协作学习过程的视频、音频、日志、生理和自我报告等多模态数据，其中的每一模态数据均有与学习投入度呈相关关系的特征变量。研究结果表明，基于多模态数据的测评能较全面地反映学生的学习投入状态，为协作学习中教师精准施策提供依据，促进教育评价数据化、科学化、精准化。
【英文摘要】：
Learning engagement is an important factor affecting students' academic success and education and teaching quality in collaborative learning. Evaluating students' learning engagement effectively can provide the evidence for teachers to analyze and judge collaborative activities. The development of artificial intelligence and intelligent perception technology leads to the smart classroom, and the evaluation method of learning engagement has also changed from a single-dimensional one to multimodal fusion analysis. Based on multimodal learning analysis theory, combined with the characteristics of collaborative learning in smart classroom, this paper discusses the deep integration of technology and learning evaluation, analyzes learning engagement from three dimensions of behavior, cognition and emotion, constructs an analysis model of smart classroom collaborative learning engagement, and applies it in practical classes. It is found that multimodal data such as video, audio, log, physiology and self-report from the collaborative learning process in smart classroom, each modal data has characteristic variables that are correlated with learning engagement. The research results show that the evaluation based on multimodal data can comprehensively reflect students' learning engagement, provide a basis for teachers to make precise policy-making in collaborative learning, and promote data-based, scientific and accurate educational evaluation.